Weather Forecasting Using Back Propagation Feed Forward Neural Network and Multiple Linear Regression

Abstract

Weather forecast is one of the most important research areas in world problems such as meteorology, human civilization, drought, agriculture and dams. We propose Back Propagation Feed Forward Neural Network and Multiple Linear Regression method and models for predicting air precipitation in the project. Proposed Neural Network model were trained with 8 optimization algorithms in order to find proper accuracy. The performance of the models in the project is evaluated with the most appropriate statistical methods. Coefficient of correlation (r), Root Means Square Error (RMSE), Mean Percent Error (MPE), Mean Absolute Percent Error (MAPE), Mean Square Error (MSE) and R-square statistics were used to measure the accuracy of the model proposed in the project. The data sets used in the project were taken from the Istanbul Provincial Meteorology System. Obtained results demonstrated that, It is seen that the model proposed in the project gives better results than the algorithms in other studies. It is seen that the model proposed in the project gives better results than the algorithms, models and techniques in other studies.